Pass more Copy types by value.

[lcnr]

There are a lot of locations where we pass &T where T: Copy by reference,
which should both be slightly less performant and less readable IMO.

This PR currently consists of three fairly self contained commits:

• passes ty::Predicate by value and stops depending on AsRef<ty::Predicate>.
• changes <&List<_>>::into_iter to iterate over the elements by value. This would break Lists
  of non copy types. But as the only list constructor requires T to be copy anyways, I think
  the improved readability is worth this potential future restriction.
• passes mir::PlaceElem by value. Mir currently has quite a few copy types which are passed by reference, e.g. Local. As I don't have a lot of experience working with MIR, I mostly did this to get some feedback from people who use MIR more frequently
• tries to reuse ty::Predicate in case it did not change in some places, which should hopefully
  fix the regression caused by Intern predicates #72055

r? @nikomatsakis for the first commit, which continues the work of #72055 and makes adding PredicateKind::ForAll slightly more pleasant. Feel free to reassign though

[lcnr]

Somewhat related, I think it might be a good idea to change Binder to also require T: Copy and to move the reference inwards where this is not the case, i.e. &Binder(T) -> Binder(&T).

[nikomatsakis]

The changes all seem reasonable to me. I guess cc @oli-obk or @eddyb on the final MIR commit, but it doesn't seem likely to be especially controversial. We've been slowly migrating over in the compiler to making more and more things Copy and, as we do, I think it makes sense to be shifting away from taking things via reference and towards taking them by value.

[lcnr]

Added one more commit here which somewhat builds on the first one.

This hopefully fixes most of the regression introduces in #72055

[Dylan-DPC-zz]

@bors try @rust-timer queue

[rust-timer]

Awaiting bors try build completion

[bors]

⌛ Trying commit f15e4b3 with merge 493b2bfa42db10f469c41cd86a5a3dc9921299f2...

[bors]

☀️ Try build successful - checks-azure
Build commit: 493b2bfa42db10f469c41cd86a5a3dc9921299f2 (493b2bfa42db10f469c41cd86a5a3dc9921299f2)

[rust-timer]

Queued 493b2bfa42db10f469c41cd86a5a3dc9921299f2 with parent 75b0a68, future comparison URL.

[eddyb]

changes List::into_iter to iterate over the elements by value.

This makes sense to me, although it's impl IntoIterator for &List<T> not for List<T>, right?
So it's <&List<_>>::into_iter not List::into_iter.

Still, I can it see it being nicer overall.

[rust-timer]

Finished benchmarking try commit 493b2bfa42db10f469c41cd86a5a3dc9921299f2, comparison URL.

[lcnr]

The above comparison link does not work for me 😅

Tried to compare it to the current master, which should still be fairly representative: https://perf.rust-lang.org/compare.html?start=3137f8e2d141d7d7c65040a718a9193f50e1282e&end=493b2bfa42db10f469c41cd86a5a3dc9921299f2&stat=instructions%3Au

This fixes slightly less than a third of the previous regression :/

[lcnr]

If I didn't miss something, this PR should be ready for merge.

[nikomatsakis]

@bors r+

[bors]

📌 Commit f15e4b3 has been approved by nikomatsakis

[lcnr]

@nikomatsakis This PR changes perf, so we probably should use @bors rollup=never here

[bors]

@lcnr: 🔑 Insufficient privileges: not in try users

[nikomatsakis]

@bors rollup=never

[Dylan-DPC-zz]

@bors p=1

[bors]

⌛ Testing commit f15e4b3 with merge 4512721...

[bors]

☀️ Test successful - checks-azure
Approved by: nikomatsakis
Pushing 4512721 to master...

[nnethercote]

Perf results from the landing show a small win. Nice work!